Pavit Chhabra

Building/Studying Cool Stuff.
SDE@Cisco
Docker Manager:

[Pavit Chhabra @mesosan16]

@cneuralnetwork Hey! I am building Bundl, it's an app that allows you to find people willing to Order from the same platform as you nearby, so that you guys can Pool your order and get that sweet offer or free delivery.

[Pavit Chhabra @mesosan16]

@bzogrammer The expansion must not stop

[S V @open_parens]

C++ should simply declare bankruptcy at this point and start again.

[Pavit Chhabra @mesosan16]

This is so insanely cool bruh wtf: https://t.co/6H86PzClIU

[OS Dev @OSdev_]

For years, people assumed pulling the power plug instantly erased everything in RAM. It doesn't.

That simple assumption led to one of the most influential hardware security papers ever published: Cold Boot Attacks.

The key observation was that DRAM doesn't lose its contents immediately after power is removed. It stores data as electrical charge in tiny capacitors, and those charges leak away gradually. At room temperature, memory can retain data for several seconds. Cool the DIMMs, and that window can extend significantly.

Researchers showed that an attacker with physical access could exploit this behavior by quickly rebooting a machine into a small custom OS or even transplanting the DIMMs into another compatible system, and dumping memory before the contents fully decayed.

Why is that useful?

Because RAM often contains things you never want an attacker to see:
• AES disk encryption keys
• RSA private keys
• Kernel secrets
• User credentials
• Session keys

The clever part wasn't just dumping memory.

Even after some bits had decayed, the researchers developed reconstruction techniques that recovered valid cryptographic keys from imperfect memory images. They successfully demonstrated attacks against systems using BitLocker, FileVault, dm-crypt, and TrueCrypt (in configurations where the encryption keys were already resident in RAM).

The paper changed an important security assumption.

The cryptography wasn't broken.
The operating system wasn't broken.

The weakness came from a physical property of DRAM itself.

It's a reminder that computer security isn't just about software bugs or weak algorithms. Sometimes, the behavior of the hardware underneath becomes part of the attack surface.

[Arika @Arikalabs]

in 2011, a researcher registered domains one bit away from popular ones not human typos, but the kind a flipped DRAM bit would produce and waited.
thousands of real machines connected.

computers were resolving DNS through corrupted memory and landing on his servers by accident.
google later found DRAM error rates in production were orders of magnitude higher than lab estimates, with roughly 8% of DIMMs seeing at least one correctable error each year.
servers have ECC. the laptop you do your banking on probably doesn't.

memory is less deterministic than most software assumes.

[V12 @v12sec]

And here's postgres bidirectional RCE

no admin required, client infects server, server infects client ♻️🐛

[Pavit Chhabra @mesosan16]

@neogoose_btw I mean.. Duh. Are you gonna get your trust store offer the network lol, how do you trust the trust store then xD

[Pavit Chhabra @mesosan16]

Oof but understandable
https://t.co/GlE3SBIJz2

[datavorous @datavorous_]

I reverse engineered Qualcomm's NPU compiler to find undocumented behaviour that affects every edge AI deployment.

Things nobody knew:

1. The compiler silently downgrades the precision of your model weights without telling you
2. Memory placement uses HiGHS which is an LP solver (not heuristics)
3. The same model on two different chips with identical reported VTCM can have 33x difference in DDR traffic
4. There's an undocumented internal simulator called Hextimate pricing ops without the hardware

Every NPU vendor be it Qualcomm, MediaTek or Apple NEVER tells you how to make the most use of their hardware.
I was very close to rage quitting before I finally lost all hope and thought of reverse engineering to understand how NPUs are handled.

Read the full write-up below:

[trash @trashh_dev]

were back to this

[Pavit Chhabra @mesosan16]

@vaxryy something's not getting reaped 😔

[Pavit Chhabra @mesosan16]

All but a human being, a moment is all I see, then I'm out 🙂

[vx-underground @vxunderground]

Literally the worst cable management I've ever seen in my life

[Pavit Chhabra @mesosan16]

6. It stops working 😔

[Strace @straceX]

Stages of lock-free programming:

1. it works
2. it works on x86
3. you learn x86 was hiding your ordering bugs the whole time
4. the ARM port ships
5. you finally read what acquire and release actually promise

[Arika @Arikalabs]

Me: std::move(x)
C++: Cool.

Me: So it's moved?
C++: I didn't say that.

[Pavit Chhabra @mesosan16]

@NGupta15442 @adonis_singh Liked lol

[haro.7z @haro7z]

I tried Vibecoding, and despite having no prior UEFI development experience, I completed the porting of edk2 to rk3326 entirely using AI.
https://t.co/jqJARC3jkz

[venture anthropologist @0xBalloonLover]

anthropic won't let you use fable for biology, chemistry, ai research, or anything that accelerates human progress. that makes it the perfect tool for developing blockchains

[Tulip King 🌷 @tulipking]

i look forward to our chinese brothers liberating the knowledge from within fable-5 and selling it to me at 5% the cost & 2x the speed